Developmental Dynamics and Driving Factors of Understory Vegetation: A Case Study of Three Typical Plantations in the Loess Plateau of China

Abstract

Understory vegetation is one of the most important links for improving forest biodiversity, and its restoration is conducive to sustainable forest development, energy flow, and nutrient cycling. However, little is known about the developmental dynamics and main driving factors of the long-time series coverage, biomass, diversity, and species composition of plantation understory vegetation. In a case study of three typical plantations, with a natural secondary forest as reference in the Loess Plateau of China, we collected understory vegetation from a Robinia pseudoacacia Linn. deciduous broad-leaved plantation, Pinus tabulaeformis Carr. evergreen coniferous plantation, and mixed plantation with an age span of 10 to 50 years. (1) The understory plantation coverage and biomass results of stands with different ages showed the R. pseudoacacia plantation to be significantly higher than the P. tabulaeformis plantation, and the species diversity of the P. tabulaeformis plantation changed the most with the stand age. However, the growth resource imbalance, and drastic changes in the stands’ environment caused by excessive intraspecific competition in the early stage of the P. tabulaeformis plantation vegetation restoration, are the main reasons that make the species diversity of undergrowth vegetation of P. tabulaeformis plantation lower than that of other stand types. (2) The understory species composition of the plantations revealed their degree of community stability. Compared to the R. pseudoacacia plantation and P. tabulaeformis plantation, the mixed plantation had higher stability, and its species composition closely resembled a natural secondary forest. The community stability of the P. tabulaeformis plantation was the lowest because it had the lowest coverage, biomass, and species diversity of understory vegetation. However, the understory species composition of the three plantation types converged, which was due to atypical species contribution. (3) The dynamic changes of canopy and soil nutrients were the main driving factors affecting the R. pseudoacacia plantation understory vegetation species composition. Stand density and elevation limited the understory vegetation communities of P. tabulaeformis plantation restoration. Soil bulk density is the key factor affecting understory vegetation in mixed plantations, and this effect weakens with the stand age. In future studies, the focus should be on the converged action and further development trend of atypical species, choosing an appropriate recovery strategy (active or passive), and providing more possibilities for the intensive management of vegetation under different plantations.

1. Introduction

One of the objectives of sustainable development is forest ecosystem protection and restoration. In recent years, this has gradually shifted from increased forest area to improved biodiversity [1,2]. Understory vegetation only represents a small part of forest biomass and is often ignored, but this does not prevent it from being integral to forest ecosystems [3,4], which also represents the majority of forest vegetation species diversity, including plantations, and contributes greatly to nutrient cycling and energy flow [3,5], and forest biodiversity is largely a function of the understory vegetation community.

The complex driving factors behind understory vegetation biomass and diversity change result from responses to resource availability [6]. This may vary depending on the forest type [7,8], age [9,10], land use type [11], and the interaction between plant species and different life forms [12,13]. These driving factors involve the different responses of understory vegetation to available light resources and soil conditions, including nutrients, moisture, and niche heterogeneity [6,14,15,16]. In addition, afforestation and plantation management can improve the ecological environment and accelerate understory vegetation restoration [17,18,19]. Therefore, plantation biodiversity improvement or decline in the Loess Plateau has recently become a focal point [18,20].

Due to its special geographical location, complex topography with vertical and horizontal gullies, and extreme climatic conditions (such as low precipitation and its significant spatiotemporal differences), the Loess Plateau has become the area most seriously affected by soil erosion in China [21]. Therefore, large-scale artificial afforestation is used for soil and water conservation, vegetation restoration, and ecological and environmental reconstruction [22]. Tree species of Robinia pseudoacacia Linn. and Pinus tabulaeformis Carr. are dominant species used in afforestation because of their ability to tolerate barren and dry environments on the Loess Plateau [23,24], and partially mature plantations have reached the generative phase [18,24]. However, since the 1960s, most Loess Plateau plantations have become mature forests, and are facing a series of ecological problems, including low understory biodiversity, slow natural succession, and unbalanced stand structure [23,25]. The undergrowth vegetation significantly affects the regeneration layer of the plantations, and changes the density, regeneration speed, and composition of woody by creating gap differences [23]. On a macroscale, elevation, aspect, and slope strongly affect the formation of understory vegetation, mainly by controlling the distribution of ecological factors such as light, heat, water, and soil nutrients [26,27,28]. On a microscale, the high variability in forest microhabitats may form asymmetric niche and competition patterns, resulting in the distribution and differentiation of understory vegetation [29]. Within a given climatic region, at the stand scale, the succession and regeneration of understory vegetation is affected by a variety of biotic and abiotic factors [30,31]. Stand structure can determine the microclimatic conditions, and affect species richness and understory regeneration [32]. Soil and topographical factors specifically affect vegetation community assembly processes through the hydrological and nutrient cycles, as well as microbial community changes [11,33]. Moreover, stand age has always been an important factor, especially regarding light heterogeneity [34]. Tsai et al. [35] showed that light heterogeneity has a positive impact on understory species diversity because it provides a diversified niche for other species with different light needs. Correspondingly, the difference in soil properties caused by forest structure and habitat condition also greatly affects understory plant species composition and interspecific relationship [36]. Therefore, restoring species diversity in plantations, and explaining the potential mechanisms and driving factors affecting forest species diversity, stability, and succession is crucial [6].

Forest management has an obvious and strong impact on stand levels because it strongly affects forest structure and composition. For forests all over the world, especially plantations, canopy coverage and stand density are the main factors driving understory vegetation change [37,38]. However, many studies do not consider the interactions between different driving factors, as well as the relative importance of quantitative driving factors, which leads to great uncertainties and differences in their evaluation and contribution. Therefore, this study uses plantations with different stand types and ages in the Loess Plateau gully region. It takes understory vegetation coverage, biomass, diversity, and species composition as the variables. It integrates environmental factors, such as topography and stand structure, to fill the gap in understanding stability and management measures. Finally, it seeks to (1) determine the trend and stability differences of understory vegetation with a stand age of typical coniferous, broad-leaved, and mixed plantation; (2) at the stand scale, analyze the differences among understory community composition and stability in different plantations from the perspective of environmental driving factors (topography, stand structure, and soil); and (3) formulate corresponding understory vegetation restoration strategies according to the compositional and stability status corresponding to stand age, and guided by existing problems and potential developmental risks.

2. Materials and Methods

2.1. Study Area

The study area is in the Caijiachuan watershed of Ji County, Shanxi Province, which is the Loess Plateau of China (Figure 1, 36°14′–36°18′ N, 110°39′–110°47′ E). This site is a typical gully area and 38 km² in size. The watershed flows from west to east, with a length of about 14 km and an elevation of 897–1557 m. The long-term mean annual air temperature is 10.2 °C, the annual precipitation is 575.9 mm, and about 70% of the precipitation occurs between June and September every year. The average annual potential evapotranspiration (PET) is 1724 mm, which far exceeds the rainfall. The soil type is primarily Haplic Luvisol (soil classification of the UN Food and Agriculture Organization [FAO]) and mostly alkaline [25]. The nested watershed forest cover rate is 75% and mainly includes R. pseudoacacia deciduous broad-leaved plantation, P. tabulaeformis evergreen coniferous plantation, and natural secondary forest dominated by temperate tree species (e.g., Quercus wutaishansea Mary, Populus simonii Carr., and Koelreuteria paniculata Laxm.). Since afforestation, most of the plantations occur on abandoned farmland, and the area has mainly regenerated naturally, logging is prohibited, and without manual intervention such as thinning and pruning.

2.2. Stand and Site Selection

A total of 48 sites were selected according to stand type and age and included a variety of structural and topographic factors to fully evaluate the impact of a variety of environmental factors on understory vegetation. These sites included two pure single tree species plantations (i.e., R. pseudoacacia (RP, Figure 2) and P. tabulaeformis (PP, Figure 2)), a mixed plantation (MP, Figure 2), and a natural secondary forest (SF, Figure 2). One hundred and seven understory vegetation species were investigated in all the sites. The basic information on all the 48 sites is given in

Table 1. Basic information related to each stand type in this study.

Stand Type	Dominant Species	Forest Age	Quantity	D (cm) 1	H (m) 2	Tree Density (Trees·ha−1)	Slope (°)	Elevation (m)
RP	Robinia pseudoacacia Linn.	10–20	3	8.94 ± 1.31	7.06 ± 1.87	800–1200	15–27	1020–1086
		20–30	3	10.19 ± 0.91	8.49 ± 1.30	1225–1700	13–34	950–1180
		30–40	3	13.25 ± 0.85	10.33 ± 0.96	1300–1525	19–33	970–1250
		40–50	3	14.49 ± 0.19	10.33 ± 1.37	900–1500	12–23	943–1218
PP	Pinus tabulaeformis Carr.	10–20	3	9.74 ± 0.44	8.44 ± 1.29	3100–5900	31–39	1064–1203
		20–30	3	11.66 ± 0.68	8.74 ± 1.05	1300–3800	22–31	1090–1357
		30–40	3	13.93 ± 0.65	8.85 ± 1.96	1100–2900	17–31	1130–1182
		40–50	3	16.86 ± 1.28	8.97 ± 1.46	1600–2100	12–25	1084–1200
MP	Robinia pseudoacacia Linn., Pinus tabulaeformis Carr., Platycladus orientalis (Linn.) Franco	10–20	3	7.22 ± 0.69	6.43 ± 0.42	975–2400	11–19	1069–1093
		20–30	3	11.32 ± 0.51	7.91 ± 0.49	1025–1725	16–27	1163–1205
		30–40	3	12.19 ± 0.48	8.32 ± 0.48	900–1725	16–26	1160–1220
		40–50	3	15.20 ± 1.26	9.15 ± 1.25	700–1175	11–22	1057–1200
SF	Quercus wutaishansea Mary, Populus simonii Carr., Koelreuteria paniculata Laxm., etc.	–	12	12.75 ± 2.19	8.58 ± 1.81	1100–1800	5–42	950–1230

¹: D indicates diameters at breast height, and the values in the table are the average value ± standard deviation. ²: H indicates the height of trees, and the values in the table are the average value ± standard deviation. RP: Robinia pseudoacacia Linn. plantation. PP: Pinus tabulaeformis Carr. plantation. MP: Mixed plantation. SF: Natural secondary forest.

.

2.3. Field Measurements and Investigation of Environmental Factors

In each stand, we randomly allocated a 20 m × 20 m sample plot to represent the stand. Spacing between stands was more than 200 m, and each stand was at least 100 m away from the edge of farmland and roads. Within a plot, the living trees with D (diameters at breast height) ≥5 cm were counted according to tree species, and all other vegetation species were defined as understory vegetation (including shrubs, herbs, and vines). All trees (D ≥ 5 cm) in a sample plot were measured. Three 2 × 2 m subplots were randomly allocated to investigate the understory vegetation (i.e., height, coverage, numbers, and species), and each subplot’s aboveground biomass was estimated using the harvest method [39]. The method to count the undergrowth vegetation used the concepts of minimal area and species-area curves by Kent [40]. Species coverage was estimated according to Kuchler [41], and a summation of the coverage of various plant layers and individual species was performed. Species-specific subplot level percent coverage data were averaged to represent the sample stand [42].

The understory vegetation and its environmental factors were investigated during the peak vegetation coverage period, namely from June to August 2021. Among them, environmental factors were divided into topographic, stand structure, and soil factors. Topographic factors included elevation (EL), slope aspect (AS), slope position (SP), and slope (S). Structure factors included average tree height (H), average diameter at breast height (D), the average height of the first live branch (LLC), stand density (SD), the average of the longest crown diameter in the N → S and E → W directions (WC), spacing coefficient (SC), average crown extension (DCE), sum of cross-sectional area at breast height per unit area (AB), and average crown length rate (RC). Among them, SC, DCE, AB, and RC are composite stand structure indices. SC represents the ratio of the average distance between trees, reflects the effective space underneath the stand [43], and was calculated as $\mathrm{S}\mathrm{C}=100/(\sqrt{\mathrm{S}\mathrm{D}}\times \mathrm{W}\mathrm{C})$. DCE reflects the outward stretching degree of the crown and was calculated as $\mathrm{D}\mathrm{C}\mathrm{E}=\mathrm{W}\mathrm{C}/\mathrm{H}$. AB was calculated as $\mathrm{A}\mathrm{B}=\mathrm{S}\mathrm{D}\times \mathsf{\pi }\times D^2/4$. RC reflects the coordination degree of tree crown and height growth and was calculated as $\mathrm{R}\mathrm{C}=$ LC/H, where LC is the average crown diameter. Soil factors included soil organic carbon (SOC), available potassium (AK), total nitrogen (TN), total phosphorus (TP), potential of hydrogen (pH), and soil bulk density (SBD). SOC was determined using the potassium dichromate method [44]. TN and TP were determined using an Automatic Chemical Analyzer (Smartchem 450, AMS Alliance, Pairs, France). AK was obtained via flame photometry (FP640). SBD was measured by the core method [45]. Soil pH was measured in a suspension using an electronic pH-meter (soil-to-water ratio of 1:2.5). Soil texture was assayed using the Mastersizer 2000 method (Malvern Mastersizer 2000, Worcestershire, UK) [46].

2.4. Data Analysis

Understory vegetation coverage is the sum of the coverage of all understory species in each site. Understory vegetation biomass is the average dry weight per unit area in each site. Species richness was the number of unique species [47]. The Shannon–Wiener index was calculated by the proportional percentage coverage of the constituent species within each sample plot and reflects the local plant community diversity [48]. The Simpson index is the probability that two individuals sampled at random in the community belong to the same species [49]. Pielou’s index was calculated based on the Simpson and the Shannon Weiner indices and reflects the degree of uniformity of individual number distribution among species [50].

To test whether understory species composition was affected by stand type and age, we conducted Permutation Multiple Analysis of Variance (PerMANOVA) on all understory vegetation. Hereafter, we used the Bray–Curtis dissimilarity matrix to summarize species composition and 999 permutations to determine statistical significance. Finally, the Non-Metric Multidimensional Scaling (NMDS) and Bray–Curtis dissimilarity measures were used to visualize the data. All calculations were performed in the R statistical program using the “vegan” package [51].

Detrended Correspondence Analysis (DCA) is considered a pre-analysis for selecting a linear or unimodal model. The results showed that the longest gradient length was less than four, so we selected the linear model, i.e., redundancy analysis (RDA), thus analyzing the relationship between understory vegetation communities and environmental factors [52]. The RDA axes were evaluated statistically by 999 Monte Carlo displacement tests. Furthermore, before performing the RDA analysis, a logarithmic conversion was performed based on the above-mentioned environmental variables (such as elevation and slope) since there are no assumptions of normality in regression on the explanatory variables [11,53]. Some non-quantitative environmental factors, such as slope position and aspect, were modified to approximate the normal distribution [53,54,55]. The analysis results of RDA were shown by a double sequence diagram. The environmental factors are generally represented by arrows. The quadrant of the arrow represents the positive and negative correlation between the environmental factors and the axes. The length of the arrow line represents the degree of correlation between an environmental factor and community distribution (the longer the line, the greater the correlation, and vice versa). The angle between the arrow line and the axes represents the correlation between an environmental factor and the axes (the smaller the angle, the higher the correlation). The angle between communities and environmental factors represents their positive and negative correlation (acute angle: positive correlation; obtuse angle: negative correlation; right angle: no correlation). Make a vertical line from different community samples to each environmental factor. The closer the projection point is, the more similar the attribute value of the environmental factor between samples is, that is, the impact of environmental factors on samples is equal.

Indicator Species Analysis (ISA) was used to analyze species that characterized stands to distinguish between stand types [56,57]. The Indicator Value (IndVals) of each species was calculated according to ISA, and its significance was confirmed by the Monte Carlo method. Moreover, the IndVals produced a combination of relative abundance (RA) and relative frequency (RF) for species in each community. ISA was performed using PC-ORD 5.0 [58]. Finally, we regard species with IndVals ≥ 0.50 in different stand types and ages as typical species.

M. Godron’s method was adopted to assess the understory community stability [59]. First, use the scatter plot formed by the percentage of understory species and the cumulative relative frequency. Then, curve fitting was performed on the scatter plot to obtain the curvilinear equation, and the intersection coordinate was calculated with the linear equation and the curvilinear equation. Finally, the linear distance between the intersection coordinate and the stable coordinate (20, 80) was calculated. The smaller the linear distance between the intersection coordinate and the stable coordinate, the more stable the understory communities. Curvilinear equation (Equation (1), from curve fitting model) and linear equation (Equation (2)) are as follows:

$$Y=a{X}^2+bX+c$$

(1)

$$Y=100-X$$

(2)

Then, substitute Equation (2) into Equation (1) to obtain intersection coordinate (M (Equation (3)), N (Equation (4))):

$$\mathrm{M}=\frac{-(b+1)\pm \sqrt{{(b+1)}^2-4a(c-100)}}{2a}$$

(3)

$$N=100-M$$

(4)

The intersection coordinate has two solutions. One solution is far greater than 100, and the other solution should be between 0 and 100. According to the research, the M and N should be greater than 0 and less than 100, so the second solution should be used.

Two-way analysis of variance (two-way ANOVA) was used to identify the effects of different stand types and age on understory vegetation coverage, biomass, and species diversity index. Pearson correlation analysis was used to analyze the correlation between environment factors. Two-way ANOVA and Pearson correlation analysis were carried out using IBM SPSS Statistics version 24. All graphics in this study were completed by OriginPro, version 2022.

3. Result

3.1. Understory Vegetation Coverage and Species Diversity

Stand age significantly impacted on understory vegetation coverage (p < 0.05,

Table 2. Two-way analysis of variance in understory vegetation coverage, biomass, and species diversity index of four stands.

Source		Type	Age	Age × Type
	df	2	3	6
Coverage	F	1.876	3.930	0.420
	p	0.175	*	0.859
Biomass	F	7.986	3.268	0.546
	p	**	*	0.768
Species richness	F	3.450	3.569	1.143
	p	*	*	0.368
Shannon–Wiener	F	3.511	3.151	1.566
	p	*	*	0.200
Pielou	F	3.506	2.790	1.464
	p	*	*	0.232
Simpson	F	3.704	2.663	1.701
	p	*	*	0.164

Note: The rows provide the degree of freedom (df), F, and p values. * p < 0.05, ** p < 0.01. SF has no stand age, and the two-way analysis of variance did not include SF.

). Stand type and age also significantly affected the species diversity index (including species richness, Shannon–Wiener, Pielou, and Simpson indices, p < 0.05). Furthermore, under different stand types, understory vegetation biomass showed highly significant differences (p < 0.01). Moreover, only stand type had no significant impact on understory vegetation coverage, and the interaction of stand type and age on understory vegetation coverage, biomass, and species diversity was also not significant.

The understory vegetation coverage and biomass of the three plantation types (including RP, PP, and MP; Figure 3) changed similarly with stand age. Stands aged between 20 to 30 years reached the lowest point, and the stand aged between 30 to 40 years plateaued. However, only the understory vegetation coverage of PP and MP with a forest age of 20 to 30 years (37.22 ± 21.50% and 39.00 ± 9.52%, respectively) was lower than SF (71.97 ± 29.58%). Additionally, the understory vegetation coverage of the three plantation types with different stand ages was greater than SF. Moreover, understory vegetation biomass was similar to the PP and MP with a forest age of 20 to 30 years (33.89 ± 26.67 g∙m⁻² and 38.06 ± 17.20 g∙m⁻², respectively), and lower than SF (46.58 ± 18.50 g∙m⁻²). Additionally, the understory vegetation biomass of the three plantation types with different stand ages was greater than SF. Furthermore, the understory vegetation coverage and biomass of RP were higher than the other three stands and reached the highest point in the stand age from 30 to 40 years (at this point in time, the coverage and biomass were 158.44 ± 27.06% and 144 ± 44.16 g∙m⁻², respectively).

The four species diversity indices were similar. PP fluctuated the most with changing stand age, while all four indices were the lowest in the 20 to 30-year-old stands, and peaked rapidly in the 30- to 40-year-old stands. Finally, when the stand age was 40 to 50 years, the values decreased slowly and were slightly higher than the stands aged 10 to 20 years (Figure 4).

The four species diversity indices for RP and PP gently decreased initially and then rose slightly with a change in stand age. However, they differed in that the inflection point of RP appeared in the stand age of 20 to 30 years, while the inflection point of MP appeared in the stand age of 30 to 40 years (Figure 4). Moreover, the four species diversity indices for SF were lower than RP and MP under different stand ages and were between the highest and lowest values for PP (Figure 4).

3.2. Species Composition

Stand type and age significantly affected the species composition of the understory vegetation communities (

Table 3. Results of the permutation multivariate analysis of variance (perMANOVA), testing the effects of stand type, stand age, and their interactions on understory species composition.

	df	F	p
Type	3	2.56	**
Age	3	1.97	**
Type × Age	9	0.97	0.679

Note: The columns provide the degree of freedom (df), F, and p values. ** p < 0.01.

). When all understory vegetation data were pooled, different stand types also occupied different partly ordination spaces (Figure 5). RP sites were grouped on the left of ordination Axis 1, PP sites were grouped on the right of ordination Axis 1 and the lower side of ordination Axis 2, and SF sites were grouped on the right of ordination Axis 1 and the upper side of ordination Axis 2. However, it is noteworthy that MP sites were grouped in the middle of ordination Axis 1 and 2, respectively, and coincided with the distribution space of the other three stand types in a large area.

When the understory vegetation components of RP, PP, and SF were separated from each other, with an increase in stand age, the RP and PP sites were close to the middle of ordination Axis 1 and 2. This is especially evident for the stands with an age of 40 to 50 years, and understory vegetation community composition increased similarly in different stands. However, the understory vegetation composition of MP under different stand ages resembled the other three stands to varying degrees.

3.3. Relationship between Understory Vegetation and Environmental Factors

According to the DCA analysis results, the maximum gradient for the first four ordination axes was less than four (

Table 4. Summary statistics for the DCA (detrended correspondence analysis).

Summary of Ordination		Axis 1	Axis 2	Axis 3	Axis 4
DCA	Eigenvalues	0.61	0.45	0.34	0.21
	Cumulative percentage variance of species data (%)	8.77	15.27	20.08	23.11
	Gradient length	3.93	3.07	3.25	2.97

). Therefore, we chose RDA to analyze the relationship between environmental factors and understory vegetation.

The multicollinearity of 22 environmental factors (including nine stand structures and four topographical factors) must be excluded before RDA. The environmental factors were, therefore, screened according to a Pearson correlation matrix (Figure 6) and stepwise regression analysis, and only the important explanatory variables were retained and used for the final collinear diagnosis (

Table 5. Multiple linear regression analysis for collinearity statistics.

	VIF	Tolerance
Constant	/	/
H	1.37	0.73
SC	1.40	0.72
RC	1.34	0.74
EL	1.53	0.66
AS	1.21	0.83
SOC	1.45	0.69
TN	1.24	0.81
TP	1.33	0.75
SBD	1.59	0.63

Note: Constant, response variables; VIF, variance inflation factor; H, the average of tree height; SC, spacing coefficient; RC, the average rate of crown length; EL, elevation; AS, slope aspect; SOC, soil organic carbon; TN, total nitrogen; TP, total phosphorus; SBD, soil bulk density.

).

Variance inflation factors (VIF) are the reciprocal of tolerance, and the degree of collinearity increases with increasing VIF. Generally, when 0 < VIF < 10, multicollinearity is absent. Therefore, no multicollinearity existed between the nine factors (Table 5), which could then be used directly for ordination analysis. Finally, three stand structures (including SC, RC, and H), two topographic factors (including EL, and AS), and four soil factors (including SOC, TN, SBD, and TP) passed collinear diagnosis and participated in RDA (Figure 7).

According to the RDA analysis (

Table 6. Summary statistics for the RDA (redundancy analysis).

Item	Axis 1	Axis 2	Axis 3	Axis 4
H	0.17	0.29 **	0.48 **	0.17
SC	0.18	0.49**	−0.11	−0.32 *
RC	0.41	−0.05 *	−0.43 **	−0.32 **
EL	0.43	−0.47 **	−0.10	−0.26
AS	0.41 **	0.00 **	0.11	0.01
SOC	0.44	0.22 **	−0.16 **	−0.12
TN	−0.64 **	0.23 *	0.03	0.06
TP	−0.09	0.18	−0.40 **	0.85 **
SBD	−0.47 **	−0.15 *	−0.57 **	−0.02 **
Eigenvalues	0.05	0.04	0.03	0.02
Cumulative percentage variance of species data (%)	4.90	8.70	11.52	13.87
Species- environment correlation	0.80	0.87	0.83	0.88
Cumulative percentage variance of species-environment	22.40	39.74	52.59	63.32
Test of significance of all canonical axes	F = 1.20                   p = 0.012

* p < 0.05, ** p < 0.01.

), the first two and four ordination axes cumulatively explained 39.74% and 63.32% of the species-environment relationship, respectively. The correlation coefficient between species-environment factors and Axis 1 was 0.80 and 0.87 for Axis 2. Moreover, Monte Carlo tests showed that the correlation between environmental and species variables was significantly represented by the first four canonical axes (F = 1.20, p = 0.012). The above results have ecological significance, and the first two ordination axes were, therefore, selected to create a two-dimensional RDA ordination diagram (Figure 7).

Ordination Axis 1 was significantly correlated with AS (r = 0.41, p < 0.01), TN (r = −0.64, p < 0.01), and SBD (r = −0.47, p < 0.01), respectively (Table 6). Also, ordination Axis 2 was significantly correlated with eight factors (p < 0.05, Table 6), but excluding TP.

From Figure 7, the understory vegetation of the four stand types with different ages had four different distribution quadrants. This shows that the understory vegetation distribution laws are restricted by various environmental conditions. RP sites are distributed mainly in the upper quadrant, with stand ages between 10 and 20 affected by stand structure factors (i.e., RC and H), and stand ages over 20 are affected by soil factors (i.e., SOC, TN, and TP). PP sites are distributed mainly in the lower right quadrant, with stand ages between 10 and 30 affected by SC most, and stand ages over 30 are affected by EL the most. MP sites are distributed mainly in the lower left quadrant, SBD significantly affecting the composition and distribution of lower stand age understory vegetation (age < 30), and with the age of the stand increased, the impact weakens accordingly. SF is randomly distributed within four quadrants, and the composition and distribution of understory vegetation are not affected by fixed or single environmental factors.

3.4. Dynamic Changes in Typical and Atypical Understory Species

The typical understory (indicator) species are shown in

Table 7. Indicator species analysis of different stand types and ages.

Type	Age	Indicator Species	Life Forms	RA	RF	IndVals	p
RP	10–20	Cirsium arvense var. integrifolium	Herb	0.79	0.83	0.66	0.001
		Wikstroemia chamaedaphne Meisn.	Shrub	0.78	0.82	0.64	0.001
		Prinsepia uniflora Batal.	Shrub	0.66	0.67	0.44	0.004
	20–30	Thladiantha dubia Bunge	Herbaceous vine	1	0.73	0.73	0.001
		Pistacia chinensis Bunge	Tree seedling	1	0.67	0.67	0.001
		Duchesnea indica (Andr.) Focke	Herb	0.81	0.67	0.54	0.002
	30–40	Ambrosia artemisiifolia L.	Herb	1	0.74	0.74	0.001
		Asparagus cochinchinensis (Lour.) Merr.	Woody vine	0.68	0.84	0.57	0.004
	40–50	Digitaria sanguinalis (L.) Scop.	Herb	1	0.83	0.83	0.001
		Ipomoea nil (Linnaeus) Roth	Voluble herb	1	0.76	0.76	0.001
		Wisteria sinensis (Sims) Sweet	Woody vine	0.77	0.67	0.52	0.002
PP	10–20	Pilea sinofasciata C. J. Chen	Herb	1	0.83	0.83	0.001
		Potentilla chinensis Ser.	Herb	1	0.83	0.83	0.001
		Commelina diffusa Burm. f.	Herb	1	0.81	0.81	0.001
	20–30	Patrinia scabiosaefolia Fisch. ex Trev.	Herb	0.8	0.67	0.54	0.002
		Sambucus williamsii Hance	Shrub	0.77	0.67	0.52	0.003
	30–40	Rhodotypos scandens (Thunb.) Makino	Shrub	1	0.71	0.71	0.001
		Rosa xanthina Lindl.	Shrub	0.66	1	0.66	0.000
		Amygdalus davidiana (Carrière) de Vos ex Henry	Tree seedling	0.77	0.67	0.52	0.002
	40–50	Celtis sinensis Pers.	Tree seedling	1	0.63	0.63	0.001
		Artemisia argyi Levl. et Van	Shrub	0.74	0.67	0.50	0.011
		Inddigofera bungeana Steud.	Shrub	0.66	0.69	0.46	0.004
MP	10–20	Rubia cordifolia L.	Herb	0.74	1	0.74	0.003
		Elaeagnus pungens Thunb.	Shrub	1	0.66	0.66	0.001
		Artemisia stechmanniana Bess.	Shrub	0.56	1	0.56	0.035
	20–30	Platycladus orientalis (L.) Franco	Tree seedling	1	0.63	0.63	0.002
		Plantago depressa Willd.	Herb	1	0.53	0.53	0.002
		Ailanthus altissima (Mill.) Swingle	Tree seedling	0.74	0.67	0.50	0.005
	30–40	Betula albosinensis Burkill	Tree seedling	1	0.75	0.75	0.001
		Melilotus officinalis (L.) Pall.	Herb	0.75	0.67	0.50	0.013
	40–50	Quercus wutaishansea Mary	Tree seedling	0.82	0.67	0.55	0.003
		Conyza canadensis (L.) Cronq.	Herb	1	0.53	0.53	0.001
SF		Vicia sepium L.	Herb	0.84	0.88	0.74	0.000
		Setaria viridis (L.) Beauv.	Herb	1	0.68	0.68	0.002
		Pogonatherum crinitum (Thunb.) Kunth	Herb	0.93	0.68	0.63	0.003
		Deutzia scabra Thunb	Herb	0.78	0.81	0.63	0.001
		Sonchus wightianus DC.	Herb	0.71	0.83	0.59	0.005
		Lespedeza bicolor Turcz.	Shrub	0.71	0.8	0.57	0.014

, displaying their dynamic changes with stand type and age and desiring to reconstruct understory environmental changes through the ecological characteristics of indicators species.

Correspondingly, we also discuss the performance and changing trend in atypical species under different stands and ages. We removed the indicator species from all stands so that the remaining species were atypical species. Under different stand types and ages, the two stands contain the same numbers of atypical understory vegetation species, the following matrix was obtained (

Table 8. Repetition numbers of atypical understory species under different stand types and ages.

Type		RP				PP				MP
	Age	10–20	20–30	30–40	40–50	10–20	20–30	30–40	40–50	10–20	20–30	30–40	40–50
RP	10–20
	20–30	15
	30–40	11	18
	40–50	6	12	13
PP	10–20	3	6	5	4
	20–30	2	3	2	5	12
	30–40	7	7	11	12	7	10
	40–50	6	7	9	10	6	8	16
MP	10–20	4	8	11	12	4	2	6	6
	20–30	3	9	9	9	5	3	7	7	15
	30–40	2	4	4	5	6	6	9	9	13	15
	40–50	7	7	9	11	6	10	11	11	6	12	13
SF	SF	6	11	11	15	6	8	7	8	9	14	18	18

). It is noteworthy that under the same stand type, greater stand age differences led to lower repetition numbers of atypical understory species. This shows the general law of community succession. With increasing stand age, the composition of atypical understory species continues to change, similar to the indicator species.

4. Discussion

Our study quantified differences in understory vegetation coverage, biomass, diversity, and species composition for different plantations, and expounded the main environmental driving factors for these differences between different stand types and ages. The restoration effect of the R. pseudoacacia plantation (RP) on understory vegetation coverage and biomass was higher than other plantations (Figure 3). However, the understory vegetation species diversity of the P. tabulaeformis plantation (PP) fluctuated the most with stand age (Figure 4), which reflected the drastic environmental changes under the forest, and we propose that this relates to stand density decline as forest age increases (Table 1). Understory vegetation species in all plantations tended to be close with increasing stand age, which was particularly obvious in the mixed plantation (MP, Figure 5). These differences are closely related to environmental drivers (Figure 7). Here, we discuss these issues, focusing on the potential environmental drivers of understory vegetation coverage, biomass, species diversity, and species composition in different stand types and ages, and briefly discuss the relationship between plantation management strategies and ecosystem functions.

4.1. Potential Drivers of Changes in Understory Vegetation Coverage and Species Diversity

Understory vegetation coverage and biomass of three plantation types first decreased and then increased with varying degrees during the middle stage (20 to 30-year age class) of vegetation restoration (Figure 3). Additionally, it is noteworthy that, in Figure 3, the lines (10 to 20-year age class) start higher than natural secondary forest (SF), and with increasing stand age, the lines of PP and MP fluctuate near the average value of SF. One possible reason is that in the early stage (10 to 20-year age class) of plantation restoration, due to the shaded environment and minimal human interference, soil erosion and high-temperature damage are significantly reduced. The abandoned farmland then quickly develops into shrub and herb communities, which then greatly improves the ecological suitability underneath the forest canopy [23,60,61]. However, with increasing plantation age, under the limited soil water carrying capacity of the vegetation (SWCCV) in arid and semi-arid areas, tree growth further reduces the growing space for understory vegetation and thereby hinders the development of the understory community [62,63]. Therefore, the understory vegetation coverage and biomass for the three plantations decreased significantly in the middle of recovery (20- to 30-year age class, Figure 3). Finally, after species competition, tree renewal, and elimination, a reasonable stand density and structure enables understory vegetation coverage and biomass to recover gradually, and ultimately closely resembles SF. In contrast to the other two plantations, the lowest point of vegetation coverage and biomass under RP was higher than SF, and the highest point (at 40- to 50-year age class) was more than double that of SF. Compared with other plantations, the lowest stand density (Table 1) of RP in each age class was the main driving factor, which greatly reduced interspecific competition, and this understory vegetation had the most sufficient growth space [37].

Similar to the changing trend in coverage and biomass, species diversity indices (Species richness, Shannon–Wiener, Pielou, and Simpson index) of the four stand types also showed a trend of first declining and then increasing. However, as stand age increased, the species diversity of vegetation under the three plantation types fluctuated and gradually became unified (closely resembling SF). Moreover, the fluctuation of PP was the highest. Zhang, et al. [64] showed that stand density affects the understory vegetation diversity index by limiting soil phosphorus availability in the Loess Plateau, which is consistent with the results of this study (Table 1). It is also noteworthy that the lowest inflection point of the four species diversity indices appeared mostly in the pure plantations (RP and PP) with a stand age of 20 to 30 years, and although the lowest inflection point of MP appeared in 30 to 40 years, it was maintained at a high level. As shown by Cavard, et al. [65] and Gosselin, et al. [66], mixed tree species increase stand structural heterogeneity and habitat diversity compared to pure forests, and allow more plant, animal, or fungal communities establish. However, this also requires mixed forest understory vegetation to have a longer interspecific competition and alternating adaptation period as the ecological environment in the forest changes (mainly the change in dominant tree species).

4.2. Potential Drivers of Species Composition Differences

Both stand type and age significantly affected the species composition of understory vegetation (p < 0.01, Table 3). NMDS analysis evaluates the interrelationships between vegetation communities [67]. According to the NMDS results (Figure 5), the understory vegetation composition of RP, PP, and SF has a relatively independent distribution space (especially for low-age stands). In contrast, the MP site occupies the area close to the center of the NMDS ordination diagram (Figure 5) and coincides with the distributional area of the three other stand types, which is consistent with the results discussed in Section 4.1. The diverse environment of a mixed forest enables the understory vegetation of other pure forests and natural secondary forests to survive and develop. A common law also exists among the three plantation stands (RP, PP, and MP), which is that as stand age increases, the understory vegetation composition unifies (the high-age stand sites move toward the center in Figure 5).

As stand age grew, the change in stand microhabitat significantly affected the understory vegetation composition [68,69]. During the growth of RP, the significant change in canopy affected the ecosystem temperature and light quantity, which is one of the main driving factors for species compositional change under the stand [70,71]. In this study, the above two conclusions were further confirmed according to the changes in indicator species under the forest (Table 7). The superior light conditions in the early stage of vegetation restoration enabled the growth of many pioneer species that require high lighting and are drought tolerant, e.g., Wikstroemia chamaedaphne Meisn. and Prinsepia uniflora Batal. With the canopy density increase, the survival rate of shade-loving species increased [72], and shade-tolerant species (e.g., Pistacia chinensis Bunge and Digitaria sanguinalis (L.) Scop.) gradually occupied the central position in the understory vegetation communities. Woody vines promote forest regeneration through direct competition with trees [73,74]. With a further increase in stand crown and shrub density, the limitation of soil nutrients (SOC, TN, and TP) on understory species is further enhanced. To compete for limited light and absorb nutrients, woody vines (e.g., Asparagus cochinchinensis (Lour.) Merr. and Wisteria sinensis (Sims) Sweet) begin to appear in the late stage of RP.

The stand density of PP in the Loess Plateau often affects soil properties, thus changing the diversity and composition of understory vegetation [64,75]. The response of the shrub layer to stand density is usually higher than the herb layer [75,76]. The density of PP in this study area generally decreases with increasing stand age (Table 1). Excessive forest density and fierce intraspecific competition are the main driving factors in the earlier stage (stand age ≤ 30 limited by SC, Figure 7). Therefore, we note that the indicator species are mainly herbs (e.g., Pilea sinofasciata C. J. Chen and Potentilla chinensis Ser.) in the early stage (stand age ≤ 20) of vegetation restoration, while shrubs (e.g., Rhodotypos scandens (Thunb.) Makino and Artemisia argyi Levl. et Van) and tree seedlings (e.g., Amygdalus davidiana (Carrière) de Vos ex Henry and Celtis sinensis Pers.) occur mainly in the middle and late stage (stand age > 20) of vegetation restoration (Table 7). Combined with the high fluctuation in understory vegetation species diversity of PP in Section 4.1, with the sudden stand density decline, the difference in the microclimates (e.g., heat radiation, air humidity, and temperature) in the forest caused by EL has become the new main driving factor (Figure 7) [33].

The understory vegetation diversity of MP in the study area is significantly higher than that of a single plantation. Gong et al. [77] showed that the main reason was that the response rate of understory vegetation diversity increases with increasing stand age. Due to niche differentiation and rotational cycle differences, multispecies forest structure often affects renewal by affecting light, water, nutrient use efficiency, and forest heterogeneity [78,79]. The multispecies MP structure usually leads to forest light heterogeneity, resulting in the coexistence of shade-intolerant and tolerant species [80,81]. We found an interesting phenomenon: the indicator species under MP implied the process of community succession. With increased stand age, it gradually changed first from light-loving pioneer shrubs (e.g., Elaeagnus pungens Thunb. and Artemisia stechmanniana Bess.) to light-loving tree seedlings (e.g., Platycladus orientalis (L.) Franco, Ailanthus altissima (Mill.) Swingle, and Betula albosinensis Burkill), and finally to moisture and water sensitive tree seedlings (Quercus wutaishansea Mary) at an age of 40 to 50 years (Table 7). Additionally, it is emphasized here that Q. wutaishansea is the main tree species in SF (Table 1). Therefore, we deduced that the understory vegetation diversity and species composition of MP not only indicate future understory species composition but also the direction of species competition among trees (this indicates that there may be more numbers and species of trees from SF coming into MP). The understory environment of MP has, therefore, obviously gradually developed into a habitat for some of the main tree species of SF. Therefore, we believe that it will further approach the developmental direction of SF.

The research of Zhang et al. [82] on SF in the Loess Plateau shows that the percentage of Fabaceae, Poaceae, and Asteraceae vegetation increases annually, corresponding to the increase in soil C, N, and P, respectively. For instance, Fabaceae vegetation increases the soil nitrogen content during succession due to its nitrogen fixation ability [83]. Poaceae vegetation fixes carbon in the atmosphere and returns it to the soil in the form of litter [84]. The six indicator species under SF in the study area included two Fabaceae vegetation (i.e., Vicia sepium L. and Lespedeza bicolor Turcz.), two Poaceae vegetation (i.e., Setaria viridis (L.) Beauv. and Pogonatherum crinitum (Thunb.) Kunth), one Asteraceae (i.e., Sonchus wightianus DC.), and one Saxifragaceae (i.e., Deutzia scabra Thunb).

Another noteworthy point is that, with increasing stand age, the repetition quantity of atypical understory species of different stand types also increased significantly (Table 7). This finding complements our previous explanation of the NMDS ordination analysis. The high-age stands in the center of Figure 5 mainly result from the contribution of atypical understory species. In particular, for MP, this is one of the main reasons why it occupies the central area in the NMDS ordination diagram. More importantly, from the perspective of the whole watershed, the convergence trend of atypical understory species in different stands shows a positive and stable signal.

Finally, by comparing plantation stands and SF, under different stand ages, the repetition numbers of MP and SF atypical understory species were the highest, while PP was the opposite. This again confirms that MP understory vegetation species composition is closer to SF.

4.3. Dynamic Response of Species Composition to Environmental Drivers

Our study found that the composition of RP understory vegetation was most affected by crown development, as the age of the stand increased, it gradually shifted to competition for soil nutrients. RP as the typical deciduous broad-leaved forest in north China, low density provides more differences and possibilities for canopy growth, and many studies deal with canopy closure [85], or gap dynamics [86], at forest stand level demonstrated that forest canopy structure significantly affects the diversity and productivity from understory vegetation, and affected understory vegetation by changing the understory light conditions [87] and through litter and nutrient cycling [88]. In addition, numerous studies have shown that phosphorus (P) is the main limiting factor for its sustainable development, and show an increasing demand for P over time [89,90], which is consistent with the results of community succession and indicator species change under the RP. Topographical and stand structure factors (here referring to the SC and EL, respectively) have a significant impact on PP understory vegetation composition. Together with this, the SC mainly affects the understory vegetation of low stand age (10 to 30-year age) PP, while the EL mainly affects high stand age (30- to 50-year age). From Table 1, the density of PP in the early stage of afforestation is much larger than the afforestation target range in the Loess Plateau (1100–1600 trees∙ha⁻¹ [91]). The effective space between trees has become the biggest factor limiting the understory vegetation species composition, which not only affects light conditions in the vertical space but also affects competition for soil nutrients and water in the horizontal space in coniferous forests [92,93]. In the middle and late stage of PP restoration, the topographical factor (i.e., EL) has the greatest impact on the understory vegetation, which is related to temperature, drought sensitivity, temporal and spatial precipitation distribution characteristics, and human activities all affect the movement of PP to a higher EL. For MP, in the early afforestation stage, it is mainly affected by SBD, and changes in SBD have the greatest influence on the spatial distribution of water and the water use strategy of vegetation in the slope [94]; with an improvement in SBD, this impact further weakens. Finally, as the reference group of this study, SF is widely distributed in the study area. Together with this, among the 12 SF sites, the understory vegetation composition is affected by multiple environmental factors, which are caused by differences in stand species composition, stand structure, soil heterogeneity, and topographical diversity. Therefore, environmental impact factors are not discussed here.

4.4. Communities Stability and Stand Management Strategy

The stability of forest vegetation communities includes constancy, resistance, resilience, and persistence [95]. This multi-dimensional comparison makes the evaluation of community stability more complex and inconvenient. Therefore, the mathematical ecological stability method (i.e., M. Godron method) that can reflect the quantity and frequency information of all species to quantitatively the stability of different stand types and ages was adopted (

Table 9. Stability analysis results under different stand types and ages (M. Godron method).

Type	Age	Curvilinear Equation	R2	Intersection Coordinates	Distance from Stable Coordinate
RP	10–20	Y = −0.0034x2 + 1.2856x	0.9336	(47.04, 52.96)	38.25
	20–30	y = −0.0044x2 + 1.4547x + 0.6436	0.8724	(43.94, 56.06)	33.85
	30–40	y = −0.0039x2 + 1.5265x + 2.6453	0.7726	(41.15, 58.85)	29.91
	40–50	y = −0.0029x2 + 1.3988x + 1.6356	0.8252	(43.27, 56.73)	32.91
PP	10–20	y = −0.0012x2 + 0.9745x + 2.5479	0.8219	(50.93, 49.07)	43.74
	20–30	y = −0.0029x2 + 1.0425x + 0.3453	0.8439	(52.74, 47.26)	46.30
	30–40	y = −0.0034x2 + 0.9851x + 6.4530	0.8942	(51.70, 48.30)	44.84
	40–50	y = −0.0085x2 + 1.2250x + 7.6346	0.7719	(51.74, 48.26)	44.89
MP	10–20	y = −0.0261x2 + 2.2244x + 1.6323	0.8369	(42.75, 57.25)	32.17
	20–30	y = −0.0181x2 + 1.9647x + 8.3525	0.7992	(41.35, 58.65)	30.20
	30–40	y = −0.0072x2 + 1.6371x + 5.8543	0.8017	(40.09, 59.91)	28.41
	40–50	y = −0.0068x2 + 1.5971x + 6.8878	0.9067	(40.05, 59.95)	28.36
SF	SF	y = −0.0207x2 + 2.3547x + 1.7453	0.7339	(38.38, 61.62)	25.99

Note: R² means coefficient of determination; stable coordinate is (20, 80).

). The smaller distance between the intersection coordinate and the stable coordinate indicates a more stable community.

Our research shows that compared with SF, the coverage, biomass, and diversity of understory vegetation of RP for different stand ages are significantly improved. However, from the perspective of community stability, there is still a big gap between RP and SF. When the stand age > 40, the community stability under RP showed a slow declining trend (Table 9). The reasons for this appearance are as follows. As a typical deciduous broad-leaved forest in the Loess Plateau, the understory vegetation species composition of RP is greatly affected by the canopy. Light availability and P element are key factors for the survival of understory vegetation and seedlings of R. pseudoacacia [89,90,96]. As the canopy closes, they may be replaced by more shade-tolerant species due to light restriction; an indicator species change confirmed this phenomenon under RP (Table 7). Végh and Tsuyuzaki [97] held that stand spatial structure is more important than stand age, and passive restoration often leads to poor communities. Since the light environment under the stand changes with canopy closure, it is valid to consider that the emergence of various shade-tolerant climbing species, here referred to as woody vines (e.g., A. cochinchinensis and W. sinensis) and voluble herbs (e.g., Ipomoea nil (Linnaeus) Roth), as indicators in the later stage of RP restoration is a potential risk hindering healthy plantation development [98]. Therefore, active canopy management is an important condition for restoring the high-quality habitat of RP. In the early stage, while preventing the height of climbing vegetation from exceeding the canopy height, expanding the canopy gap can effectively prevent the growth of shade-tolerant climbing species [98,99]; In the later stage, corresponding soil nutrient cycling elements, especially soil phosphorus can be supplemented [89,90], which can promote the growth of young trees under the forest, as well as improving species diversity.

For PP, its community stability is the worst among all stand types and has not changed significantly with the increase in stand ages (Table 9). The drastic change in the understory environment is one of the main reasons for community instability. Understory species change caused by sudden stand density decline (Table 1) is at the expense of species diversity to some extent, which greatly reduces the stability of resistance of understory communities. Therefore, there is poor resistance stability (juvenile and mid-aged stands) under PP and a severe fluctuation of vegetation diversity under different stand ages (Figure 4). Expanding the effective distance between trees in the early stage of restoration is an important prerequisite for reducing resource competition and promoting the stability and succession efficiency of understory communities [64,100,101]. Concurrently, in the near-mature forest at a later restoration stage (age > 30), attention should be paid to the influence of elevation on the composition of understory vegetation, here mainly referring to the possible positive or negative effects of population migration phenomena. On the basin scale in arid areas, great differences exist in the distribution of temperature, moisture, and soil nutrients, resulting in differences in vegetation growth and stability [102].

Additionally, according to the NMDS (Figure 5), the two pure plantations with single tree species differed greatly during the early stage, but species compositional changes under the forest also increased with increasing stand age, and a unified trend existed near the “central area”. Also, for MP, the understory vegetation for different stand ages was more obviously closer to this “central area”. More importantly, while the understory vegetation of MP maintained a higher community stability and its community stability closer to SF (Table 9), Q. wutaishansea, as the main tree species of SF, became the main indicator species of high-age MP, which undoubtedly implies the succession direction of MP understory vegetation and the developmental direction of tree competition. However, for MP in the early stages of recovery, SBD reflecting the maturity of soil under the stand was the dominant factor limiting the allocation strategy of vegetation growth resources (especially water use) [94,101]. Therefore, minimizing unnecessary human disturbance is a necessary condition for the succession of vegetation under MP to approach and eventually reach near natural secondary forest.

5. Conclusions

To reduce the negative impact of plantations on understory community stability and succession, the relationship between the environmental drivers limiting its development and understory vegetation should first be understood. Generally, the coverage, biomass, species diversity, and composition of understory vegetation vary with stand type, age, and environment factors. In this study, canopy and soil nutrient factors are the main driving forces affecting the understory species composition and stability of the R. pseudoacacia plantation (RP). For the P. tabulaeformis plantation (PP), topographical and stand structure factors (i.e., spacing coefficient and elevation) are the main restrictive factors for understory vegetation development in medium and young stands (age < 30) and near mature stands (age > 30), respectively, with low community stability. Additionally, the understory community stability of the mixed plantation (MP) is higher than the two pure plantations with single tree species, and the species composition more closely resembles natural secondary forest. Soil bulk density is the main restrictive factor for understory vegetation development in medium and young stands (age < 30). For the understory vegetation of two pure forests, we, therefore, prefer an active restoration strategy (e.g., expanding the canopy gap, supplementing soil nutrients, and increasing the effective distance under the stand). Together with this, for MP, a passive restoration strategy (e.g., reducing human interference) is preferred. Furthermore, future research should focus on the converged action and further development trend of atypical species, and provide more possibilities for the intensive management of vegetation under different plantations. This study is a case of three typical plantations in the Loess Plateau of China. Through the analysis and comparison of environmental factors affecting the abundance, diversity, and species composition of understory vegetation in plantations and natural secondary forests, we aim to guide the sustainable development and management of the understory vegetation of forests on a worldwide scale and expect to accelerate the development of vegetation communities to a more stable and advanced stage.